Will AI come for your job?

How do I open my two-day workshop?

What can I possibly say that will motivate a very senior, training-fatigued audience to care about why we are here? Why should they pay attention?

Here’s the thing. I’ve been arguing for years that pharma needs to change the way it operationalizes its commercial approach. It was, in fact, my raison d’être for starting my company. And based on what I observe when working with my clients, most companies are still stuck in an outdated model. Despite my best efforts, and those of many like-minded consultants and employees.

But now something has happened. Artificial intelligence has broken out of its cage and is roaming free. It will have consequences.

I was wrong about AI

My initial thought, as the wave started rolling over us all a couple of years ago, was that pharma would still need a field force. AI won’t be able to do what a human can do. The physical interaction, the human-to-human engagement, will still matter. Perhaps even more as AI evolves.

I have since recalibrated.

As I see it now, AI is the thing that will kill pharma’s current commercial model. And when it does, it will happen fast and a lot of people will be out of a job.

Let me tell you why that is.

What the doctors are already doing

Let’s look at some recent data on AI adoption among specialist physicians. An international study from March 2026 surveyed 1,165 physicians across 15 specialties in seven countries (US, UK, Canada, Germany, France, Italy, China).

92% use generative AI in clinical practice. Not a single specialty falls below 85%.

And what do they use it for? Reviewing clinical guidelines. Summarizing study data. Comparing drugs within the same class. Real-time support in treatment decisions. Answers to complex patient cases.

Read that list again. That’s our arena. That’s the exact content of a pharma rep’s call deck and an MSL’s scientific exchange.

But wait, there’s more, as they used to say on the commercial morning TV-shows: 80% are satisfied with the AI-generated answers to their medical questions. 73% say AI influences their clinical decisions, 12% significantly and 61% to some degree. And 85% expect to increase their use further in the coming year.

And this is the situation today. Think about two, three years from now. Do you think these numbers will go down?

The conclusion this points to

For me, all of this points to one thing:

The value of information exchange, between pharma and healthcare, will approach zero.

What do I mean by that? The current commercial model is built on representatives, from both sales and medical, sharing the latest information about their products. And that sharing is done in a way that provides a one-sided argument for why this particular product should be used. Doctors and nurses are not stupid, they know the information is biased. But up until now they have been willing to accept that, because they have seen value in this exchange.

But that is about to end.

The OpenEvidence effect

OpenEvidence, often described as “ChatGPT for doctors”, is an AI platform that answers clinical questions with sourced, referenced summaries of the medical literature, built on content partnerships with journals like NEJM and JAMA. It’s free for verified clinicians.

The adoption numbers are staggering. By spring 2026, OpenEvidence reported daily use by more than 40% of the US physician workforce, with over 20 million clinical consultations per month. On March 10, 2026, it passed one million consultations between verified physicians and an AI system in a single day. A first in history.

Think about what that means. A physician with a clinical question no longer needs to wait for the next rep visit or the next symposium. She gets a referenced, unbiased answer in seconds. From a source that has no product to sell.

Meanwhile, the door keeps closing

We have already watched access to individual decision makers approach zero. Today, pharma reps mostly present to groups busy eating the free lunch provided.

At the same time, the data tells us this: only 45% of healthcare professionals are open to meetings with pharma, down from 60% just eighteen months earlier. And of those who do take meetings, fewer than 1 in 7 feel that representatives truly understand their needs.

All of this should make pharma executives question everything they know about how to sell pharmaceuticals. Something needs to change. Now.

The way forward

So what is the way forward?

Here’s some data that answers two questions at once: What has healthcare always struggled with? And what is it that AI cannot do?

Healthcare’s oldest problem is not lack of knowledge. It’s implementation. It takes on average 17 years for new clinical knowledge to reach patient benefit. Printed educational materials, the classic “here’s the latest data” approach, improve clinical practice by a median of 4.3% according to a Cochrane review. Information alone changes almost nothing.

And AI does not escape this problem. It suffers from it. 80% of AI projects in healthcare never make it past the pilot stage. The barriers are workflows, medical record systems, governance and change management.

So what actually works? Human support. A meta-analysis of 23 studies covering 1,398 care units found that units receiving practical, recurring facilitation from a person were nearly three times more likely to adopt evidence-based practice (95% CI 2.18 to 3.43).

Information doesn’t implement itself. AI doesn’t implement itself. People implement. Together with other people.

In essence

Pharma must abandon transactional meetings based on information exchange to value-based partnerships focused on helping healthcare systems implement innovative treatments.

I know there are great examples of this already happening out there. I’ve seen it with my own eyes, as the result of my own efforts. But there is so much more to be done.

Your move

If you, dear reader, feel AI may impact your job, start learning how to generate a different kind of value for your customers. Don’t wait for your manager to get you going, because he or she is equally stuck in the traditional model.

Don’t wait for anyone.

Start today if you want to keep your job.

There you have it. All of a sudden, I now know how to open my two-day workshop – a couple of slides showing what the future holds, a future that should provide plenty enough motivation to sit through two days with a maintained focus. Because what we will be doing together is practising what a different conversation will look like. A conversation that is of equal interest and value to both you and your customer.

And if you read this far, I think you ready for that conversation as well.

/Mats

Background research & sources

• International specialist physician study, March 2026 (n=1,165, 15 specialties, 7 countries), referenced in EMARKETER, June 2026. Most used tools: ChatGPT 83%, Gemini 50%, OpenEvidence 36%, UpToDate with AI search 32%.
• OpenEvidence company announcements and press coverage, 2026: 40%+ of US physicians as daily users, 1 million consultations in a single day (March 10, 2026).
• Veeva Pulse Field Trends, 2024; PrescriberPoint, 2024: HCP openness to pharma meetings 45%, down from 60%; fewer than 1 in 7 feel understood.
• Balas & Boren / Morris et al.: 17-year average lag from clinical knowledge to patient benefit.
• Cochrane review, Printed educational materials: +4.3% median effect on clinical practice.
• Healthcare AI implementation reports, 2026: ~80% of healthcare AI projects stall at pilot stage.
• Läkarförbundet AI survey, 2024: 6% of Swedish physicians’ workplaces have AI guidelines.
• Baskerville et al., Annals of Family Medicine 2012: practice facilitation meta-analysis, 23 studies, 1,398 care units, OR 2.76 (95% CI 2.18–3.43).